Mapping Bus and Stream Travel Time Using Machine Learning Approaches

Table 4

Comparison of prediction accuracy between proposed and existing approaches in terms of MAPE.

Section	Proposed Approaches		Existing approaches
	GBM	SVM	Linear regression	ANN
TCS to VN	18.54	20.17	24.53	22.10
VN to TCS	23.14	24.14	35.58	28.54
LM to Indira Nagar	11.41	23.03	38.45	24.27
Indira Nagar to CSIR	20.30	22.63	27.51	24.01

The values in bold indicate the lowest error among the four proposed methods for each of the study corridor. It is intended to show the best performing model. For example, the value 18.54 in bold represents that GBM method performs better than SVM, LR, and ANN approaches in TCS to VN link.